Image processing model training method, device, equipment, medium and program product

Abstract

The application relates to an image processing model training method and device, equipment, a medium and a program product. The method comprises the following steps: extracting a first feature map from an original noise image, performing noise reduction processing based on the first feature map to obtain a noise reduction image; extracting a second feature map from the noise reduction image, performing image reconstruction based on the second feature map to obtain a reconstructed noise image; obtaining an expected noise-free image corresponding to the original noise image, determining an adversarial loss between the noise reduction image and the expected noise-free image; determining a reconstruction loss between the reconstructed noise image and the original noise image, performing model training based on the adversarial loss and the reconstruction loss, and obtaining an image processing model. The method can effectively improve the signal-to-noise ratio, reduce background interference, and optimize the balance between imaging speed and image quality.